Five Chinese models you can call for free through one OpenAI-compatible endpoint. 10 free-model calls a day, 5 requests per minute, no credit card, no deposit, no phone number.
Get a free API key Verify the limits| Model | Type | Context | Notes |
|---|---|---|---|
glm-4.7-flash | Text | 202K | Zhipu's production flash tier, good for everyday chat and code |
glm-4.1v-thinking-flash | Text (reasoning) | 128K | Visible reasoning steps |
glm-4.6v-flash | Text + vision | — | Send an image URL or base64 in the messages array |
cogview-3-flash | Image generation | — | Text-to-image, returns image URLs |
cogvideox-flash | Video generation | — | Text-to-video, capped at 5 free generations/day |
curl https://aiapi-pro.com/v1/chat/completions \
-H "Authorization: Bearer $NOVAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"glm-4.7-flash","messages":[{"role":"user","content":"Hello"}]}'Any OpenAI SDK works: set base_url to https://aiapi-pro.com/v1 and the model id to one of the five above. Keys start with nvai-.
These are the providers' own Flash tiers served through official endpoints — not demos, and not a time-limited promotion. The free tier has a daily call cap; the models themselves are the production Flash models. There is no markup on paid usage either: the gateway passes through provider list price with a 1.0 multiplier.
Model choice is usually where the bill comes from, so the free tier is meant to let you measure before you spend. If you hit the daily cap: a one-time top-up removes it and raises the limit to 60 requests/minute; the first top-up carries a 25% bonus. Paid Chinese models start around $0.02 per 1M tokens for the cheapest text models — the live rate card is at /china-ai-leaderboard.json.
Yes. Five models are free to call: glm-4.7-flash, glm-4.6v-flash, glm-4.1v-thinking-flash, cogview-3-flash (image) and cogvideox-flash (video). You need an email account and an API key, but no credit card, deposit or phone number.
10 free-model calls per day and 5 requests per minute on the free tier, resetting at 00:00 UTC+8. Free video generations are capped at 5 per day. Paid models are separate and optional.
Requests to free models return a 429 until the next reset. A one-time top-up removes the free-tier cap and unlocks 60 requests per minute; the first top-up also carries a 25% bonus.
They are the providers' own Flash tiers, not demos: GLM-4.7-Flash and GLM-4.1V-Thinking-Flash are Zhipu's production flash models, CogView3-Flash is an image model and CogVideoX-Flash is a video model. Larger flagship models are available on the paid catalog.
Not sure what a model id means? Decode it here.